0

I have 3 remote VMs. All have both public and internal IP. All blong to eth0. I have installed Hadoop 3.1.1 on them. For master node, I wrote its internal ip as master host in its config files (core-site, yarn-site). and public ip of other VMs in its wokers file. Fo wokers, I wrote public ip in all config files. but it does not work and jobs stuck in accepted status waiting for container to be lunched!!! It seems that the workers want to connect with public ip but master is listening on internal one!!!! If i set public ip for master VM, obviously it does not work! If i set interal ip of master in workers machine, resourcemanager shows them as CORRUPT datanodes; So there is no active datanode!! Any Idea how to set IPs? Thanks in advance

Ebrahim Salehi
  • 21
  • 4
  • Start with all IPs being externally facing. Clients need to address datanodes to read and write, not only the resource manager. Also, look at the job logs to know why they're only accepted. Is it waiting for resources or a network connection? – OneCricketeer Aug 28 '18 at 11:38
  • Thanks, it was security issue – Ebrahim Salehi Aug 29 '18 at 11:27

0 Answers0